Researchers at the Oxford Internet Institute have published a study projecting ‘the future accumulation of profiles belonging to deceased Facebook users’. Carl Öhman and David Watson used the social network’s ‘audience insights’ data, which businesses use to target their adverts, to find out how many ‘monthly active users’ of different ages there are across the world, and combined this with life expectancy data to create their models. If Facebook continues to grow at its current rate, by 2100 it will host the accounts of 4.9 billion dead people.

That’s an upper limit; the lower limit – if Facebook acquires no new users between now and the end of the century – is 1.4 billion, ‘fully 98 per cent of the 1.43 billion users in our dataset’. The researchers concede that ‘both scenarios are implausible’ – the actual figure will be somewhere between the two extremes – but they show that ‘Facebook will indubitably have hundreds of millions of dead users by 2060 if not sooner.’

Nobody knows how many dead users Facebook has now; probably not even Facebook. It won’t publish official figures (despite being constantly embroiled in privacy scandals, it’s very guarded about the data it releases). When someone dies, their friends or family can report it and have the account ‘memorialised’. You can appoint someone as a ‘legacy contact’ to manage your account after your death, or you can choose to have your account deleted by going to your ‘memorialisation settings’ and selecting ‘delete after death’. Many accounts, however, are neither deleted nor memorialised, but left more or less as they were: ghosts in the network.

Facebook’s algorithms blithely carry on including the dead in ‘highlights of the year’ videos, and suggesting them as invitees for parties. In April, Sheryl Sandberg announced that the company is actively working to get ‘faster and better’ at detecting dead accounts to stop this happening, but admitted it was not straightforward.

Different lives leave different patterns of grief. I asked my friends about the ways they interact with Facebook’s ghosts. Some said they quickly deleted the contact to protect themselves from unexpected reminders of the dead person; some said they still liked, commented on and shared old photos, the same way they always had; some said they posted commemorative messages on significant dates; one person said they deleted their own account to avoid a ghost with a lot of shared digital history. It’s unclear how Facebook could use these patterns of behaviour – hard to distinguish from the ways people interact with the profiles of the living – to identify dead users.

It isn’t only users’ life-or-death status that Facebook can misrepresent. On a micro level, almost nothing on my Facebook page is a true reflection of my life: some events and relationships are over documented, and everything else is ignored. On a macro level, Facebook boasted to advertisers in 2017 that it had 41 million users aged between 18 and 24 in the United States; according to the US census, there were only 31 million people in that age group.

There are different ways of defining numbers of ‘monthly active users’ but Facebook’s is undeniably generous, including anyone who is logged in once during the month. Accounts are configured by default to ‘keep me signed in’, so many people who rarely use it are almost never logged out. Some websites automatically disconnect users after periods of inactivity but Facebook never does, and accounts are left signed in on its mobile app even if the phone dies.

Facebook accounts, Öhman and Watson say, are ‘part of our shared cultural digital heritage, which may prove invaluable not only to future historians, but to future generations’. I think of my own account and hope no one in the future ever takes it in earnest, or views it as a guide to the way we really lived. Perhaps a more useful discussion would be about the faith we collectively place in Facebook as a record keeper now. Or, as Öhman and Watson put it, their results ‘should be interpreted not as a prediction of the future, but as a commentary on the present’.